First an intro. The alias Cafinux is because my name is too common to use on most forums and Cafinux best describes me as a user of coffee and Linux in quantity.
I’m enjoying running a version 3.04 server for a few cameras which are recording some modifications to my home. So congrats you have created an excellent application here. Only problem I have had was with the install scripts for buster, but I see it is noted already. Obviously I’m just schedule recording but I would love to do a time lapse as well. Yes I have a python script somewhere, to do it but wondered if you were intending to add the feature?
We haven’t talked about this feature…:when I think time lapsed I think construction cameras that record a days worth of events in 30 minutes…but I could be wrong.
Can you tell me more about what you are thinking, what features it needs and typically the ratio of actual recording to minutes the average time lapse video would need to be?
Yes you are right the construction industry does use them. They use them for a number of reasons, not just security, but as a way to insist on health and safety requirements by sub contractors, time keeping, weather, etc. The files are vastly reduced from the full recording and can be used for PR in the finished job. I think they started with the PR thing with Government money spending, but found them useful for a host of other reasons. They were impressed with how sparse the frame rate could be but retain good information.
The script I played with could just rip off a series of recording files taking one frame in so many, but I think you could do it by time too.
Interesting.
What are the average video lengths for a 8 hour or 24 hour video? How many frames are typically saved per hour or minute?
I’ll do some finding out.
Research
A note before I start. Adding this capability to Bluecherry might not be the best option, and running it as a separate operation could be a better one. I was surprised no one else hadn’t asked about it, as it is quite a powerful tool. Also to note is that quite a lot of cameras can supply a second stream as a time lapse. But it is really only useful for quick viewing, as they are normally of a reduced quality and don’t stand up in a court of law.
The main reason that a time lapse is used is for is reduced file size. On the surface it may appear to be not the case, but when you dig into it, file size is the issue.
Of all the scenarios I studied, the time between frames is governed by how fine grained the observation needed to be. Someone only wanting proof something was done on a given day can have very large frame spacing. On the other hand a surprising amount can happen in 3 seconds if the actual carrying out of an action is needed to be observed. Whether it is a machine been operated or an area that someone can pass through quickly.
So I found that there were two distinct frame rate groups. The higher frame rate, parts of a second to up to about 10 seconds. The top end of this is normally used for promotion and advertising. The other group frame rates are measured in minutes. One to over 30 minutes, quite a big spread on this group. My study was on a very small sample of user’s, so could easily be misleading in some way. I noticed that there was a notable concentration around the one second mark but getting the information, which is normally confidential isn’t easy.
i would like to see a time-lapse recording of a set of cameras:
have a grid layout of some cameras and take a 4k snapshot of that grid of cameras once per second and compile it into a video. that way you could review everything that happened at night in about 30 min. my boss would pay a thousand bucks for the implementation of that feature if you are interested.
A couple questions. $1000 may not cover all of the development, we have to implement entirely new rules in the server for handling image files, new schedule based on ‘time lapsed’, handle storage of tmp files…etc. But it’s worth more discussion to find out:
-
Do you need the server to do this time lapse compiling automatically (12 hours blocks) or when a user selects a time frame and the server would then process the job?
-
Would you need this for each camera, or only specific cameras?
-
Do you need video saved for the entire duration or just image files every second?
-
Do we need to keep this video file when it is finished building, or delete it?
we have 5 outdoor cameras and 15 indoor cameras. spider webs and wind tend to make motion detection almost useless. the boss arrives in the morning and runs the client on his computer on a dedicated display and then proceeds to review each camera feed in like 16x speed through all the motion trigger videos but that takes a lot of his time. what he would like to do is view a single video file that has all the cameras or a select set of cameras in grid view (essentially just as it appears on his monitor in live view) in time-lapse for the hours 6 pm to 4 am and the whole day sunday. if there is an issue he can pause the video at a certain time and review a particular time on a particular camera’s video file. if there is no obvious issue he could move on with his day. from what I understand there is d.v.r software out there that does this already but I forget which one.
i don’t think he would be willing to pay more than $1000 for the development of this feature. he is more likely to have me set up a windows d.v.r on a the maintenance server for that purpose with the other d.v.r software. however, if you can tell me how much it would take to make it worth your while I can relay that amount to him.
Do you need the server to do this time lapse compiling automatically (12 hours blocks) or when a user selects a time frame and the server would then process the job?
either would work but it would be better to set a time frame with a specific set of cameras.
Do we need to keep this video file when it is finished building, or delete it?
it sounds like you would have the video compile from the recordings of select cameras. i was assuming you would simply take a low-rez snapshot of all the selected devices’ live streams and assemble the snapshots into a single image in a grid pattern and then use those grid images to compile a gif-like video.
regardless of your method, once the video has been viewed it becomes useless. if you were to do it your way it could be deleted upon closing, if you were to do it my way you could have it auto-delete once the file is over 60 hours old.
Would you need this for each camera, or only specific cameras?
time lapse would only be for sets of cameras into a single file video file with them all playing concurrently side by side in a single video file. it would be just as if he were watching the grid of live cameras only in time-lapse for a selected or scheduled time frame (depending on how you choose to implement it).
Do you need video saved for the entire duration or just image files every second?
I’m not sure what this means but maybe my other answers and descriptions have already answered this question. if not, please restate the question.